Figure 2.

Scatter plots of the average variance in a network and the pairwise agreement between hierarchies from different ranking metrics. The average variance is calculated as the mean of the variances of the estimated treatment effects and describes the average information present in a network. More imprecise network are on the right-hand side of the plots. Spearman $\rho$ (top row) and Average Overlap (bottom row) values for the pairwise agreement between $p_{BV}$ and SUCRA_F (first column), SUCRA_F and relative treatment effect (second column),$p_{BV}$ and relative treatment effect (third column). Purple line: cubic smoothing spline with five degrees of freedom. p _BV, probability of producing the best value; SUCRA_F, surface under the cumulative ranking curve (calculated in frequentist setting).